A direct recap of Google’s Search Central Live Deep Dive in Barcelona reports unusually specific timing estimates for discovery, crawling and indexing. The figures are useful for setting expectations, but their most consequential message is variability. A typical processing time cannot tell a publisher when a particular page will enter Search or become eligible for an AI answer.
What was reported in Barcelona
In ROAST’s October 2, 2026 first-hand recap, John Campbell attributes the figures to Gary Illyes and Google’s internal analysis. Reported typical times include approximately 20 hours for new-URL discovery, 30 days for refreshing a known URL, 24 hours for sitemap processing and 1.5 hours for end-to-end indexing. Canonicalisation changes are listed at one to three weeks and site moves at one to three months.
The same recap gives core-update recovery as typically three to six months, with slower cases extending to six months or a year. Some crawling and indexing processes have a slowest outcome of “never,” with quality mentioned as a factor. It also notes that stages depend on one another, so delays can accumulate. These are event-reported figures; the recap does not provide a published Google methodology sufficient to reproduce the analysis or identify what statistical measure “typical” represents.
Discovery is not indexing
A newly discovered URL is not necessarily a fetched page, an indexed document or a ranked result. Those distinctions matter when interpreting the numbers. Adding 20 hours and 1.5 hours would not establish a universal publication-to-indexing deadline: it leaves other dependencies unresolved and assumes the reported typical values describe the same journey.
Google’s published recrawling guidance remains broader. It says crawling can take a few days to a few weeks and that requesting a crawl does not guarantee inclusion. Repeated requests for the same URL do not accelerate the process. That documentation is a useful counterweight to treating an event estimate as a service commitment.
Nor should a 30-day refresh figure be interpreted as a schedule shared by every known URL. Google’s crawling overview explains that some news homepages may be revisited every few minutes. A site’s own observed behaviour is therefore more informative for operational decisions than assigning every page the same waiting period.
Use the numbers to frame an investigation
For a newly published page, establish what is known at each stage. Check whether Google can reach the URL, whether server logs record a fetch, and what URL Inspection reports about indexing and the selected canonical. Keep publication time separate from first observed Googlebot access and the later inspection result. Otherwise, a single “not visible yet” complaint combines several different problems.
After a migration or canonical change, inspect the implementation and signals rather than waiting passively for the reported interval to expire. Equally, a delay beyond a typical value is not, by itself, evidence of a quality problem. Technical findings and content assessment need their own support.
The AI Search implication follows from Google’s eligibility guidance: a supporting page must be indexed and eligible to appear in Search with a snippet. Eligibility does not guarantee selection or citation. For time-sensitive publishers, track the progression from accessible page to indexed page separately from any observed appearance in AI answers.
These reported timings can make planning conversations more realistic. They are most useful when paired with a record of what Google has actually done on the site, rather than turned into countdowns after which indexing, migration completion or ranking recovery is assumed to be owed.